John V has posted some graphics recently arguing that CRN1-2 yielded pretty much the same results as major temperature indices and, in some sense, vindicated these results. As Gavin Schmidt has pointed out, the U.S. is only 2% of the world’s surface; and as I’ve observed on many occasions, the statistical methodologies and data quality in the U.S. are different from the ROW, as is the history.
I’ve been focussing more on ROW issues recently rather than U.S. issues, but thought that it would be interesting to drop in briefly on the U.S. issues. Since there seems to be an interest in gridding, I’ll probably do a version, using the methods that I used to contour the trend maps a while ago. I also thought that it would be interesting to verify John V’s CRN1-2 estimate to NOAA and NASA versions myself.
I mention 2006 in the header because there’s something interesting about 2006 which hasn’t been mentioned in John V’s posts: 2006 values for both NOAA and NASA are as much as 0.8 deg C higher than CRN1-2 values.
Let’s review the bidding on 2006 in the U.S. In January 2007, NOAA (not NASA) announced that 2006 was the warmest U.S. year ever:
The 2006 average annual temperature for the contiguous U.S. was the warmest on record and nearly identical to the record set in 1998, according to scientists at the NOAA National Climatic Data Center in Asheville, N.C.
In May, 2006 was demoted slightly to 2nd place, just behind 1998.
Using final quality controlled data from the recently completed USHCN Version 2 data set, the 2006 average annual temperature for the contiguous U.S. was the 2nd warmest on record.
The following graphic shows the differences between NOAA and CRN1-2 averages (the latter using John V’s numbers, which he emailed me and can be downloaded from here. ) In this graphic, there is obviously a remarkable difference between the 2006 NOAA values and the CRN1-2 values with an extraordinarily large difference of about 0.8 deg C between the NOAA index and the CRN1-2 index.
Now we knew that NOAA runs warmer than NASA – so what is the corresponding result for NASA. See below:
Figure 2. Difference between GISS and CRN1-2 (both centered on 1951-1980, the NASA reference period).
Using the CRN1-2 index, 2006 is not the 1st, 2nd or even 3rd warmest year but the 31st warmest year. The 2006 anomaly exists for both NOAA and NASA. What causes it? Worth looking into. Why didn’t this show up in John V’s graphs? Unfortunately his graphs were typically smoothed and 2006 got lost in his end-point handling, I guess. Personally with these relatively short series, I much prefer to keep an unsmoothed version at least in a grey-scale with a smooth overlay, so that this type of thing shows up better.
Note: I’ll try to do my own gridding of CRN1-2 data to verify whether the 2006 anomaly is due to some artifact in John V’s gridding calculation as opposed to an artifact in the NASA-NOAA numbers.
UPDATE: John V has observed below that the yearly data that he sent me is only up to March 2006 whereas GISS and NOAA values that have compared (in anomaly form) are for the full 2006. He also sent a monthly data set here.
In writing this post, I did not consider the possibility that the annual data for 2006 covered only the first 3 months. When John V sent me his data, he did not state that the 2006 average was only for the first 3 months (it could be deduced from the monthly data, but since the graphs presented to date were from annual data, that’s what I looked at.) This might account for the difference.
Indeed, it’s quite possible that the 2006 anomaly noticed here is an “artifact” – a possibility that I specifically mentioned prior to this update.
I’m going to do some gridding my own way and I’ll revert on this. If any misunderstanding has arisen out of the 3-month stub, I don’t think that, under the circumstances, I’m entirely responsible; in any event, we’ll clear it up; blogs can be quickly edited and I’m quite happy to re-post reconciled graphics.
Also, readers should understand that no slight is intended in the above remarks towards the integrity of John V’s analysis. In comments below, it’s suggested that I intended a slight, but was “too clever” to actually put it in words. If I wanted to make a slight, I would have; the reason why there is no slight “in words” is because there is no slight. John V’s put his methods online and sent me an ASCII version promptly when I requested.
However, different people can look at data differently. How NOAA and CRN1-2 versions compare is just as relevant to me as how GISS and CRN1-2 compare and that comparison was what initiated the present post. It was an area where I expected a difference and it was a topic that John V had not commented on.
I did not suggest that he was in any sense covering up a NOAA-CRN1-2 discrepancy by not comparing them – only that he hadn’t done so and I think that it’s worth doing and worth commenting on – as the warmest 2006 claim primarily originates from NOAA and it’s quite permissible to use CRN1-2 station information to check that result.